| Date | Content | Reading | Slides |
|---|---|---|---|
| 3/31 - 4/2 | Welcome/Overview; Maximum likelihood estimation |
Probability review: Murphy 2.1-2.4, 2.6, 2.8, 3.1-3.2 Statistics review, maximum likelihood: Murphy 4.2 |
Slides, Annotated slides |
| Supervised learning: linear models | |||
| 4/2 - 4/4 | Linear regression |
Linear algebra review: Murphy 7.1-7.3 Matrix calculus review: Murphy 7.8 Maximum likelihood regression: Murphy 4.2 Linear regression: Murphy 11-11.2 |
Slides, Annotated slides, Linear regression colab, diabetes.txt |
| 4/7 - 4/9 | Linear regression with basis functions; Cross-validation |
Maximum likelihood regression: Murphy 4.2 Linear regression: Murphy 11-11.2 |
Slides, Annotated slides, Polynomial regression colab |
| 4/9 - 4/11 | Bias-variance trade-off |
Bias-variance trade-off: Murphy 4.7.6 |
Slides, Annotated slides, Bias / variance colab |
| 4/14 - 4/18 | Regularization and sparsity, LASSO |
Ridge regression: Murphy 11.3-11.4 Lasso regression: Murphy 11.4 |
Slides, Annotated slides, Ridge and Lasso colab, house_train_kaggle.csv |
| 4/18 - 4/23 | Gradient descent |
Gradient descent: Murphy 8-8.2.1 |
Slides, Annotated slides |
| 4/23 - 4/28 | Convexity; Gradient descent analysis; Stochastic gradient descent |
Stochastic gradient descent: Murphy 8.4-8.4.4 |
Slides, Annotated slides |
| 4/28 - 4/30 | Classification; Logistic regression | Logistic regression: Murphy 10-10.2.4, 10.3-10.3.3 |
Slides, Annotated slides Classification and Logistic Regression colab |
| Fri 5/2 | Midterm | See Exams Page. | |
| 5/5 | Classification 2; prediction pitfalls |
Logistic regression: Murphy 10-10.2.4, 10.3-10.3.3 |
Slides, Annotated slides |
| Supervised learning: non-linear models | |||
| 5/7 - 5/9 | Bootstrap; Kernel methods |
Bootstrap: Efron and Hastie 10.2, 11-11.2 Kernels: Bishop 6-6.2, Murphy 17, 17.1, 17.3.4, 17.3.9 |
Slides, Annotated slides |
| 5/9 - 5/12 | Neural networks |
Neural Networks : Murphy 13-13.4.3 |
Slides, Annotated slides |
| 5/14 - 5/16 | Non-parametric methods; Nearest neighbors | Nearest neighbors: Murphy 16.1 | Slides, Annotated slides |
| 5/16 - 5/19 | More non-parametric methods; Tree-based |
Trees, Random Forrests: Murphy 18 Gradient Boosting Trees: Murphy 18 |
Slides, Annotated slides |
| Unsupervised learning | 5/21 - 5/23 | Principal component analysis (PCA) |
PCA, Singular value decomposition: Murphy 20.1 Kernel PCA: Murphy 20.4.6 |
Slides, Annotated slides |
| 5/23 - 5/28 | Singular value decomposition (SVD); more matrix decompositions; autoencoders |
Autoencoders: Murphy 20.3, 22.1 |
Slides, Annotated slides | 5/30 - 6/2 | K-means; Gaussian mixture models (GMMs) | K-means, GMM: Murphy 21.3-21.5 | Slides, Annotated slides |
| Modern machine learning (Advanced Topics) | |||
| 6/4 | Multi-armed bandits (Leo Maynard-Zhang) |
Multi-armed bandits: Bandit Algorithms textbook, Jamieson informal notes Linear bandits: linear bandits paper, generalized linear bandits paper, pure exploration/BAI paper Contextual bandits: contextual bandits survey paper |
Slides, |
| 6/6 | Reinforcement Learning for training Large Language Models |
Suggested readings for more info (not required):
Reinforcement Learning (RL): Sutton and Barto textbook, OpenAI's spinning up Large Language Models (LLMs): original transformer paper, visual explanation, wikipedia on LLMs RL for LLMs: KL-control, reward model, InstructGPT paper (ChatGPT), recent DeepSeek R1 paper |
Slides, Youtube version |